1 files changed, 190 insertions, 319 deletions

diff --git a/fs/xfs/xfs_bmap_util.c b/fs/xfs/xfs_bmap_util.c
index 541d59f5e65..64731ef3324 100644
--- a/fs/xfs/xfs_bmap_util.c
+++ b/fs/xfs/xfs_bmap_util.c
@@ -18,31 +18,31 @@
  */
 #include "xfs.h"
 #include "xfs_fs.h"
+#include "xfs_shared.h"
 #include "xfs_format.h"
+#include "xfs_log_format.h"
+#include "xfs_trans_resv.h"
 #include "xfs_bit.h"
-#include "xfs_log.h"
-#include "xfs_inum.h"
-#include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
 #include "xfs_mount.h"
-#include "xfs_da_btree.h"
-#include "xfs_bmap_btree.h"
-#include "xfs_alloc_btree.h"
-#include "xfs_ialloc_btree.h"
-#include "xfs_dinode.h"
+#include "xfs_da_format.h"
 #include "xfs_inode.h"
 #include "xfs_btree.h"
+#include "xfs_trans.h"
 #include "xfs_extfree_item.h"
 #include "xfs_alloc.h"
 #include "xfs_bmap.h"
 #include "xfs_bmap_util.h"
+#include "xfs_bmap_btree.h"
 #include "xfs_rtalloc.h"
 #include "xfs_error.h"
 #include "xfs_quota.h"
 #include "xfs_trans_space.h"
 #include "xfs_trace.h"
 #include "xfs_icache.h"
+#include "xfs_log.h"
+#include "xfs_dinode.h"
 
 /* Kernel only BMAP related definitions and functions */
 
@@ -249,48 +249,6 @@ xfs_bmap_rtalloc(
 }
 
 /*
- * Stack switching interfaces for allocation
- */
-static void
-xfs_bmapi_allocate_worker(
-	struct work_struct	*work)
-{
-	struct xfs_bmalloca	*args = container_of(work,
-						struct xfs_bmalloca, work);
-	unsigned long		pflags;
-
-	/* we are in a transaction context here */
-	current_set_flags_nested(&pflags, PF_FSTRANS);
-
-	args->result = __xfs_bmapi_allocate(args);
-	complete(args->done);
-
-	current_restore_flags_nested(&pflags, PF_FSTRANS);
-}
-
-/*
- * Some allocation requests often come in with little stack to work on. Push
- * them off to a worker thread so there is lots of stack to use. Otherwise just
- * call directly to avoid the context switch overhead here.
- */
-int
-xfs_bmapi_allocate(
-	struct xfs_bmalloca	*args)
-{
-	DECLARE_COMPLETION_ONSTACK(done);
-
-	if (!args->stack_switch)
-		return __xfs_bmapi_allocate(args);
-
-
-	args->done = &done;
-	INIT_WORK_ONSTACK(&args->work, xfs_bmapi_allocate_worker);
-	queue_work(xfs_alloc_wq, &args->work);
-	wait_for_completion(&done);
-	return args->result;
-}
-
-/*
  * Check if the endoff is outside the last extent. If so the caller will grow
  * the allocation to a stripe unit boundary.  All offsets are considered outside
  * the end of file for an empty fork, so 1 is returned in *eof in that case.
@@ -612,31 +570,32 @@ xfs_getbmap(
 
 	if (bmv->bmv_count > ULONG_MAX / sizeof(struct getbmapx))
 		return XFS_ERROR(ENOMEM);
-	out = kmem_zalloc(bmv->bmv_count * sizeof(struct getbmapx), KM_MAYFAIL);
-	if (!out) {
-		out = kmem_zalloc_large(bmv->bmv_count *
-					sizeof(struct getbmapx));
-		if (!out)
-			return XFS_ERROR(ENOMEM);
-	}
+	out = kmem_zalloc_large(bmv->bmv_count * sizeof(struct getbmapx), 0);
+	if (!out)
+		return XFS_ERROR(ENOMEM);
 
 	xfs_ilock(ip, XFS_IOLOCK_SHARED);
-	if (whichfork == XFS_DATA_FORK && !(iflags & BMV_IF_DELALLOC)) {
-		if (ip->i_delayed_blks || XFS_ISIZE(ip) > ip->i_d.di_size) {
+	if (whichfork == XFS_DATA_FORK) {
+		if (!(iflags & BMV_IF_DELALLOC) &&
+		    (ip->i_delayed_blks || XFS_ISIZE(ip) > ip->i_d.di_size)) {
 			error = -filemap_write_and_wait(VFS_I(ip)->i_mapping);
 			if (error)
 				goto out_unlock_iolock;
+
+			/*
+			 * Even after flushing the inode, there can still be
+			 * delalloc blocks on the inode beyond EOF due to
+			 * speculative preallocation.  These are not removed
+			 * until the release function is called or the inode
+			 * is inactivated.  Hence we cannot assert here that
+			 * ip->i_delayed_blks == 0.
+			 */
 		}
-		/*
-		 * even after flushing the inode, there can still be delalloc
-		 * blocks on the inode beyond EOF due to speculative
-		 * preallocation. These are not removed until the release
-		 * function is called or the inode is inactivated. Hence we
-		 * cannot assert here that ip->i_delayed_blks == 0.
-		 */
-	}
 
-	lock = xfs_ilock_map_shared(ip);
+		lock = xfs_ilock_data_map_shared(ip);
+	} else {
+		lock = xfs_ilock_attr_map_shared(ip);
+	}
 
 	/*
 	 * Don't let nex be bigger than the number of extents
@@ -741,7 +700,7 @@ xfs_getbmap(
  out_free_map:
 	kmem_free(map);
  out_unlock_ilock:
-	xfs_iunlock_map_shared(ip, lock);
+	xfs_iunlock(ip, lock);
  out_unlock_iolock:
 	xfs_iunlock(ip, XFS_IOLOCK_SHARED);
 
@@ -754,10 +713,7 @@ xfs_getbmap(
 			break;
 	}
 
-	if (is_vmalloc_addr(out))
-		kmem_free_large(out);
-	else
-		kmem_free(out);
+	kmem_free(out);
 	return error;
 }
 
@@ -972,32 +928,12 @@ xfs_free_eofblocks(
 	return error;
 }
 
-/*
- * xfs_alloc_file_space()
- *      This routine allocates disk space for the given file.
- *
- *	If alloc_type == 0, this request is for an ALLOCSP type
- *	request which will change the file size.  In this case, no
- *	DMAPI event will be generated by the call.  A TRUNCATE event
- *	will be generated later by xfs_setattr.
- *
- *	If alloc_type != 0, this request is for a RESVSP type
- *	request, and a DMAPI DM_EVENT_WRITE will be generated if the
- *	lower block boundary byte address is less than the file's
- *	length.
- *
- * RETURNS:
- *       0 on success
- *      errno on error
- *
- */
-STATIC int
+int
 xfs_alloc_file_space(
-	xfs_inode_t		*ip,
+	struct xfs_inode	*ip,
 	xfs_off_t		offset,
 	xfs_off_t		len,
-	int			alloc_type,
-	int			attr_flags)
+	int			alloc_type)
 {
 	xfs_mount_t		*mp = ip->i_mount;
 	xfs_off_t		count;
@@ -1195,9 +1131,15 @@ xfs_zero_remaining_bytes(
 	xfs_buf_unlock(bp);
 
 	for (offset = startoff; offset <= endoff; offset = lastoffset + 1) {
+		uint lock_mode;
+
 		offset_fsb = XFS_B_TO_FSBT(mp, offset);
 		nimap = 1;
+
+		lock_mode = xfs_ilock_data_map_shared(ip);
 		error = xfs_bmapi_read(ip, offset_fsb, 1, &imap, &nimap, 0);
+		xfs_iunlock(ip, lock_mode);
+
 		if (error || nimap < 1)
 			break;
 		ASSERT(imap.br_blockcount >= 1);
@@ -1214,7 +1156,12 @@ xfs_zero_remaining_bytes(
 		XFS_BUF_UNWRITE(bp);
 		XFS_BUF_READ(bp);
 		XFS_BUF_SET_ADDR(bp, xfs_fsb_to_db(ip, imap.br_startblock));
-		xfsbdstrat(mp, bp);
+
+		if (XFS_FORCED_SHUTDOWN(mp)) {
+			error = XFS_ERROR(EIO);
+			break;
+		}
+		xfs_buf_iorequest(bp);
 		error = xfs_buf_iowait(bp);
 		if (error) {
 			xfs_buf_ioerror_alert(bp,
@@ -1227,7 +1174,12 @@ xfs_zero_remaining_bytes(
 		XFS_BUF_UNDONE(bp);
 		XFS_BUF_UNREAD(bp);
 		XFS_BUF_WRITE(bp);
-		xfsbdstrat(mp, bp);
+
+		if (XFS_FORCED_SHUTDOWN(mp)) {
+			error = XFS_ERROR(EIO);
+			break;
+		}
+		xfs_buf_iorequest(bp);
 		error = xfs_buf_iowait(bp);
 		if (error) {
 			xfs_buf_ioerror_alert(bp,
@@ -1239,24 +1191,11 @@ xfs_zero_remaining_bytes(
 	return error;
 }
 
-/*
- * xfs_free_file_space()
- *      This routine frees disk space for the given file.
- *
- *	This routine is only called by xfs_change_file_space
- *	for an UNRESVSP type call.
- *
- * RETURNS:
- *       0 on success
- *      errno on error
- *
- */
-STATIC int
+int
 xfs_free_file_space(
-	xfs_inode_t		*ip,
+	struct xfs_inode	*ip,
 	xfs_off_t		offset,
-	xfs_off_t		len,
-	int			attr_flags)
+	xfs_off_t		len)
 {
 	int			committed;
 	int			done;
@@ -1274,7 +1213,6 @@ xfs_free_file_space(
 	int			rt;
 	xfs_fileoff_t		startoffset_fsb;
 	xfs_trans_t		*tp;
-	int			need_iolock = 1;
 
 	mp = ip->i_mount;
 
@@ -1291,20 +1229,15 @@ xfs_free_file_space(
 	startoffset_fsb	= XFS_B_TO_FSB(mp, offset);
 	endoffset_fsb = XFS_B_TO_FSBT(mp, offset + len);
 
-	if (attr_flags & XFS_ATTR_NOLOCK)
-		need_iolock = 0;
-	if (need_iolock) {
-		xfs_ilock(ip, XFS_IOLOCK_EXCL);
-		/* wait for the completion of any pending DIOs */
-		inode_dio_wait(VFS_I(ip));
-	}
+	/* wait for the completion of any pending DIOs */
+	inode_dio_wait(VFS_I(ip));
 
 	rounding = max_t(xfs_off_t, 1 << mp->m_sb.sb_blocklog, PAGE_CACHE_SIZE);
 	ioffset = offset & ~(rounding - 1);
 	error = -filemap_write_and_wait_range(VFS_I(ip)->i_mapping,
 					      ioffset, -1);
 	if (error)
-		goto out_unlock_iolock;
+		goto out;
 	truncate_pagecache_range(VFS_I(ip), ioffset, -1);
 
 	/*
@@ -1318,7 +1251,7 @@ xfs_free_file_space(
 		error = xfs_bmapi_read(ip, startoffset_fsb, 1,
 					&imap, &nimap, 0);
 		if (error)
-			goto out_unlock_iolock;
+			goto out;
 		ASSERT(nimap == 0 || nimap == 1);
 		if (nimap && imap.br_startblock != HOLESTARTBLOCK) {
 			xfs_daddr_t	block;
@@ -1333,7 +1266,7 @@ xfs_free_file_space(
 		error = xfs_bmapi_read(ip, endoffset_fsb - 1, 1,
 					&imap, &nimap, 0);
 		if (error)
-			goto out_unlock_iolock;
+			goto out;
 		ASSERT(nimap == 0 || nimap == 1);
 		if (nimap && imap.br_startblock != HOLESTARTBLOCK) {
 			ASSERT(imap.br_startblock != DELAYSTARTBLOCK);
@@ -1373,7 +1306,6 @@ xfs_free_file_space(
 		 * the freeing of the space succeeds at ENOSPC.
 		 */
 		tp = xfs_trans_alloc(mp, XFS_TRANS_DIOSTRAT);
-		tp->t_flags |= XFS_TRANS_RESERVE;
 		error = xfs_trans_reserve(tp, &M_RES(mp)->tr_write, resblks, 0);
 
 		/*
@@ -1419,27 +1351,23 @@ xfs_free_file_space(
 		xfs_iunlock(ip, XFS_ILOCK_EXCL);
 	}
 
- out_unlock_iolock:
-	if (need_iolock)
-		xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+ out:
 	return error;
 
  error0:
 	xfs_bmap_cancel(&free_list);
  error1:
 	xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
-	xfs_iunlock(ip, need_iolock ? (XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL) :
-		    XFS_ILOCK_EXCL);
-	return error;
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	goto out;
 }
 
 
-STATIC int
+int
 xfs_zero_file_space(
 	struct xfs_inode	*ip,
 	xfs_off_t		offset,
-	xfs_off_t		len,
-	int			attr_flags)
+	xfs_off_t		len)
 {
 	struct xfs_mount	*mp = ip->i_mount;
 	uint			granularity;
@@ -1447,6 +1375,8 @@ xfs_zero_file_space(
 	xfs_off_t		end_boundary;
 	int			error;
 
+	trace_xfs_zero_file_space(ip);
+
 	granularity = max_t(uint, 1 << mp->m_sb.sb_blocklog, PAGE_CACHE_SIZE);
 
 	/*
@@ -1460,26 +1390,32 @@ xfs_zero_file_space(
 	ASSERT(start_boundary >= offset);
 	ASSERT(end_boundary <= offset + len);
 
-	if (!(attr_flags & XFS_ATTR_NOLOCK))
-		xfs_ilock(ip, XFS_IOLOCK_EXCL);
-
 	if (start_boundary < end_boundary - 1) {
-		/* punch out the page cache over the conversion range */
+		/*
+		 * punch out delayed allocation blocks and the page cache over
+		 * the conversion range
+		 */
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		error = xfs_bmap_punch_delalloc_range(ip,
+				XFS_B_TO_FSBT(mp, start_boundary),
+				XFS_B_TO_FSB(mp, end_boundary - start_boundary));
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
 		truncate_pagecache_range(VFS_I(ip), start_boundary,
 					 end_boundary - 1);
+
 		/* convert the blocks */
 		error = xfs_alloc_file_space(ip, start_boundary,
 					end_boundary - start_boundary - 1,
-					XFS_BMAPI_PREALLOC | XFS_BMAPI_CONVERT,
-					attr_flags);
+					XFS_BMAPI_PREALLOC | XFS_BMAPI_CONVERT);
 		if (error)
-			goto out_unlock;
+			goto out;
 
 		/* We've handled the interior of the range, now for the edges */
-		if (start_boundary != offset)
+		if (start_boundary != offset) {
 			error = xfs_iozero(ip, offset, start_boundary - offset);
-		if (error)
-			goto out_unlock;
+			if (error)
+				goto out;
+		}
 
 		if (end_boundary != offset + len)
 			error = xfs_iozero(ip, end_boundary,
@@ -1493,194 +1429,103 @@ xfs_zero_file_space(
 		error = xfs_iozero(ip, offset, len);
 	}
 
-out_unlock:
-	if (!(attr_flags & XFS_ATTR_NOLOCK))
-		xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+out:
 	return error;
 
 }
 
 /*
- * xfs_change_file_space()
- *      This routine allocates or frees disk space for the given file.
- *      The user specified parameters are checked for alignment and size
- *      limitations.
- *
+ * xfs_collapse_file_space()
+ *	This routine frees disk space and shift extent for the given file.
+ *	The first thing we do is to free data blocks in the specified range
+ *	by calling xfs_free_file_space(). It would also sync dirty data
+ *	and invalidate page cache over the region on which collapse range
+ *	is working. And Shift extent records to the left to cover a hole.
  * RETURNS:
- *       0 on success
- *      errno on error
+ *	0 on success
+ *	errno on error
  *
  */
 int
-xfs_change_file_space(
-	xfs_inode_t	*ip,
-	int		cmd,
-	xfs_flock64_t	*bf,
-	xfs_off_t	offset,
-	int		attr_flags)
+xfs_collapse_file_space(
+	struct xfs_inode	*ip,
+	xfs_off_t		offset,
+	xfs_off_t		len)
 {
-	xfs_mount_t	*mp = ip->i_mount;
-	int		clrprealloc;
-	int		error;
-	xfs_fsize_t	fsize;
-	int		setprealloc;
-	xfs_off_t	startoffset;
-	xfs_trans_t	*tp;
-	struct iattr	iattr;
-
-	if (!S_ISREG(ip->i_d.di_mode))
-		return XFS_ERROR(EINVAL);
-
-	switch (bf->l_whence) {
-	case 0: /*SEEK_SET*/
-		break;
-	case 1: /*SEEK_CUR*/
-		bf->l_start += offset;
-		break;
-	case 2: /*SEEK_END*/
-		bf->l_start += XFS_ISIZE(ip);
-		break;
-	default:
-		return XFS_ERROR(EINVAL);
-	}
-
-	/*
-	 * length of <= 0 for resv/unresv/zero is invalid.  length for
-	 * alloc/free is ignored completely and we have no idea what userspace
-	 * might have set it to, so set it to zero to allow range
-	 * checks to pass.
-	 */
-	switch (cmd) {
-	case XFS_IOC_ZERO_RANGE:
-	case XFS_IOC_RESVSP:
-	case XFS_IOC_RESVSP64:
-	case XFS_IOC_UNRESVSP:
-	case XFS_IOC_UNRESVSP64:
-		if (bf->l_len <= 0)
-			return XFS_ERROR(EINVAL);
-		break;
-	default:
-		bf->l_len = 0;
-		break;
-	}
-
-	if (bf->l_start < 0 ||
-	    bf->l_start > mp->m_super->s_maxbytes ||
-	    bf->l_start + bf->l_len < 0 ||
-	    bf->l_start + bf->l_len >= mp->m_super->s_maxbytes)
-		return XFS_ERROR(EINVAL);
-
-	bf->l_whence = 0;
+	int			done = 0;
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_trans	*tp;
+	int			error;
+	xfs_extnum_t		current_ext = 0;
+	struct xfs_bmap_free	free_list;
+	xfs_fsblock_t		first_block;
+	int			committed;
+	xfs_fileoff_t		start_fsb;
+	xfs_fileoff_t		shift_fsb;
 
-	startoffset = bf->l_start;
-	fsize = XFS_ISIZE(ip);
+	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
 
-	setprealloc = clrprealloc = 0;
-	switch (cmd) {
-	case XFS_IOC_ZERO_RANGE:
-		error = xfs_zero_file_space(ip, startoffset, bf->l_len,
-						attr_flags);
-		if (error)
-			return error;
-		setprealloc = 1;
-		break;
+	trace_xfs_collapse_file_space(ip);
 
-	case XFS_IOC_RESVSP:
-	case XFS_IOC_RESVSP64:
-		error = xfs_alloc_file_space(ip, startoffset, bf->l_len,
-						XFS_BMAPI_PREALLOC, attr_flags);
-		if (error)
-			return error;
-		setprealloc = 1;
-		break;
+	start_fsb = XFS_B_TO_FSB(mp, offset + len);
+	shift_fsb = XFS_B_TO_FSB(mp, len);
 
-	case XFS_IOC_UNRESVSP:
-	case XFS_IOC_UNRESVSP64:
-		if ((error = xfs_free_file_space(ip, startoffset, bf->l_len,
-								attr_flags)))
-			return error;
-		break;
+	error = xfs_free_file_space(ip, offset, len);
+	if (error)
+		return error;
 
-	case XFS_IOC_ALLOCSP:
-	case XFS_IOC_ALLOCSP64:
-	case XFS_IOC_FREESP:
-	case XFS_IOC_FREESP64:
+	while (!error && !done) {
+		tp = xfs_trans_alloc(mp, XFS_TRANS_DIOSTRAT);
 		/*
-		 * These operations actually do IO when extending the file, but
-		 * the allocation is done seperately to the zeroing that is
-		 * done. This set of operations need to be serialised against
-		 * other IO operations, such as truncate and buffered IO. We
-		 * need to take the IOLOCK here to serialise the allocation and
-		 * zeroing IO to prevent other IOLOCK holders (e.g. getbmap,
-		 * truncate, direct IO) from racing against the transient
-		 * allocated but not written state we can have here.
+		 * We would need to reserve permanent block for transaction.
+		 * This will come into picture when after shifting extent into
+		 * hole we found that adjacent extents can be merged which
+		 * may lead to freeing of a block during record update.
 		 */
-		xfs_ilock(ip, XFS_IOLOCK_EXCL);
-		if (startoffset > fsize) {
-			error = xfs_alloc_file_space(ip, fsize,
-					startoffset - fsize, 0,
-					attr_flags | XFS_ATTR_NOLOCK);
-			if (error) {
-				xfs_iunlock(ip, XFS_IOLOCK_EXCL);
-				break;
-			}
+		error = xfs_trans_reserve(tp, &M_RES(mp)->tr_write,
+				XFS_DIOSTRAT_SPACE_RES(mp, 0), 0);
+		if (error) {
+			xfs_trans_cancel(tp, 0);
+			break;
 		}
 
-		iattr.ia_valid = ATTR_SIZE;
-		iattr.ia_size = startoffset;
-
-		error = xfs_setattr_size(ip, &iattr,
-					 attr_flags | XFS_ATTR_NOLOCK);
-		xfs_iunlock(ip, XFS_IOLOCK_EXCL);
-
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		error = xfs_trans_reserve_quota(tp, mp, ip->i_udquot,
+				ip->i_gdquot, ip->i_pdquot,
+				XFS_DIOSTRAT_SPACE_RES(mp, 0), 0,
+				XFS_QMOPT_RES_REGBLKS);
 		if (error)
-			return error;
-
-		clrprealloc = 1;
-		break;
-
-	default:
-		ASSERT(0);
-		return XFS_ERROR(EINVAL);
-	}
+			goto out;
 
-	/*
-	 * update the inode timestamp, mode, and prealloc flag bits
-	 */
-	tp = xfs_trans_alloc(mp, XFS_TRANS_WRITEID);
-	error = xfs_trans_reserve(tp, &M_RES(mp)->tr_writeid, 0, 0);
-	if (error) {
-		xfs_trans_cancel(tp, 0);
-		return error;
-	}
-
-	xfs_ilock(ip, XFS_ILOCK_EXCL);
-	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
+		xfs_trans_ijoin(tp, ip, 0);
 
-	if ((attr_flags & XFS_ATTR_DMI) == 0) {
-		ip->i_d.di_mode &= ~S_ISUID;
+		xfs_bmap_init(&free_list, &first_block);
 
 		/*
-		 * Note that we don't have to worry about mandatory
-		 * file locking being disabled here because we only
-		 * clear the S_ISGID bit if the Group execute bit is
-		 * on, but if it was on then mandatory locking wouldn't
-		 * have been enabled.
+		 * We are using the write transaction in which max 2 bmbt
+		 * updates are allowed
 		 */
-		if (ip->i_d.di_mode & S_IXGRP)
-			ip->i_d.di_mode &= ~S_ISGID;
+		error = xfs_bmap_shift_extents(tp, ip, &done, start_fsb,
+					       shift_fsb, &current_ext,
+					       &first_block, &free_list,
+					       XFS_BMAP_MAX_SHIFT_EXTENTS);
+		if (error)
+			goto out;
+
+		error = xfs_bmap_finish(&tp, &free_list, &committed);
+		if (error)
+			goto out;
 
-		xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+		error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
 	}
-	if (setprealloc)
-		ip->i_d.di_flags |= XFS_DIFLAG_PREALLOC;
-	else if (clrprealloc)
-		ip->i_d.di_flags &= ~XFS_DIFLAG_PREALLOC;
 
-	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
-	if (attr_flags & XFS_ATTR_SYNC)
-		xfs_trans_set_sync(tp);
-	return xfs_trans_commit(tp, 0);
+	return error;
+
+out:
+	xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	return error;
 }
 
 /*
@@ -1789,14 +1634,6 @@ xfs_swap_extents(
 	int		taforkblks = 0;
 	__uint64_t	tmp;
 
-	/*
-	 * We have no way of updating owner information in the BMBT blocks for
-	 * each inode on CRC enabled filesystems, so to avoid corrupting the
-	 * this metadata we simply don't allow extent swaps to occur.
-	 */
-	if (xfs_sb_version_hascrc(&mp->m_sb))
-		return XFS_ERROR(EINVAL);
-
 	tempifp = kmem_alloc(sizeof(xfs_ifork_t), KM_MAYFAIL);
 	if (!tempifp) {
 		error = XFS_ERROR(ENOMEM);
@@ -1920,6 +1757,42 @@ xfs_swap_extents(
 			goto out_trans_cancel;
 	}
 
+	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+	xfs_trans_ijoin(tp, tip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+
+	/*
+	 * Before we've swapped the forks, lets set the owners of the forks
+	 * appropriately. We have to do this as we are demand paging the btree
+	 * buffers, and so the validation done on read will expect the owner
+	 * field to be correctly set. Once we change the owners, we can swap the
+	 * inode forks.
+	 *
+	 * Note the trickiness in setting the log flags - we set the owner log
+	 * flag on the opposite inode (i.e. the inode we are setting the new
+	 * owner to be) because once we swap the forks and log that, log
+	 * recovery is going to see the fork as owned by the swapped inode,
+	 * not the pre-swapped inodes.
+	 */
+	src_log_flags = XFS_ILOG_CORE;
+	target_log_flags = XFS_ILOG_CORE;
+	if (ip->i_d.di_version == 3 &&
+	    ip->i_d.di_format == XFS_DINODE_FMT_BTREE) {
+		target_log_flags |= XFS_ILOG_DOWNER;
+		error = xfs_bmbt_change_owner(tp, ip, XFS_DATA_FORK,
+					      tip->i_ino, NULL);
+		if (error)
+			goto out_trans_cancel;
+	}
+
+	if (tip->i_d.di_version == 3 &&
+	    tip->i_d.di_format == XFS_DINODE_FMT_BTREE) {
+		src_log_flags |= XFS_ILOG_DOWNER;
+		error = xfs_bmbt_change_owner(tp, tip, XFS_DATA_FORK,
+					      ip->i_ino, NULL);
+		if (error)
+			goto out_trans_cancel;
+	}
+
 	/*
 	 * Swap the data forks of the inodes
 	 */
@@ -1957,7 +1830,6 @@ xfs_swap_extents(
 	tip->i_delayed_blks = ip->i_delayed_blks;
 	ip->i_delayed_blks = 0;
 
-	src_log_flags = XFS_ILOG_CORE;
 	switch (ip->i_d.di_format) {
 	case XFS_DINODE_FMT_EXTENTS:
 		/* If the extents fit in the inode, fix the
@@ -1971,11 +1843,12 @@ xfs_swap_extents(
 		src_log_flags |= XFS_ILOG_DEXT;
 		break;
 	case XFS_DINODE_FMT_BTREE:
+		ASSERT(ip->i_d.di_version < 3 ||
+		       (src_log_flags & XFS_ILOG_DOWNER));
 		src_log_flags |= XFS_ILOG_DBROOT;
 		break;
 	}
 
-	target_log_flags = XFS_ILOG_CORE;
 	switch (tip->i_d.di_format) {
 	case XFS_DINODE_FMT_EXTENTS:
 		/* If the extents fit in the inode, fix the
@@ -1990,13 +1863,11 @@ xfs_swap_extents(
 		break;
 	case XFS_DINODE_FMT_BTREE:
 		target_log_flags |= XFS_ILOG_DBROOT;
+		ASSERT(tip->i_d.di_version < 3 ||
+		       (target_log_flags & XFS_ILOG_DOWNER));
 		break;
 	}
 
-
-	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
-	xfs_trans_ijoin(tp, tip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
-
 	xfs_trans_log_inode(tp, ip,  src_log_flags);
 	xfs_trans_log_inode(tp, tip, target_log_flags);